Amazon's Agentic Ads Just Went Live. Make Your Catalog the Product the Assistant Recommends.

This week at Cannes, timed to Prime Day, Amazon turned on Alexa+ Agentic Ads, a format where a shopper sees an ad and finishes the purchase without leaving it, food delivery or concert tickets, all inside one conversation. Papa Johns and a Sony ticketing arm are in the beta. The format is the headline. The shift underneath it is what should change how you run Amazon this quarter.

Demand is moving into the conversation. Amazon has folded Rufus, its shopping assistant used by more than 300 million customers last year, and Alexa+ into a single Alexa for Shopping layer that answers questions, compares products, checks price history, and builds the cart. When the assistant answers, it names one or two products, not a shelf of ten. The recommendation is the new shelf. If your product is not the one it names, the impression never reaches your keyword report and the sale never carries your campaign attribution.

The trap hiding in your ROAS

Here is the part that will fool careful marketers. As more generic demand resolves inside the assistant, your reported PPC ROAS can look better while your real base shrinks. The conversation intercepts the broad, non-branded queries, your ads keep serving mostly to people already hunting your brand, and the efficiency number climbs even as you lose new-customer reach. You will not catch it in ROAS. You will catch it late, when your category rank slips.

So stop watching ROAS alone and start watching two leading indicators. First, branded versus non-branded impression share, read weekly. When non-branded softens on broad terms while branded holds, the assistant is taking the generic discovery. Second, the gap between detail-page sessions and units. If units hold or grow while page sessions fall, agent-led purchases are filling the gap and your old funnel view is going blind. These two numbers see the shift months before Best Seller Rank does.

Be the product the assistant names

The defensive move is boring and mandatory. clean catalog data, accurate price, consistent reviews. The offensive move is what most sellers have not built. When an assistant summarizes options, it leans on structured attributes and on reviews. A listing missing material, fit, use case, compatible-with, or certifications is invisible to the summarizer even if it ranks fine in classic search. A product with strong but stale reviews loses inside a chat answer to a fresher competitor faster than it would in organic results.

So the highest-return work right now is filling every attribute Amazon exposes for your category, writing detail pages that read cleanly to a machine summarizer, and keeping a steady review cadence so the assistant has fresh signal to quote. Existing sponsored campaigns already serve inside Alexa for Shopping automatically, so the lever you control is not the ad, it is whether your catalog is legible to the agent deciding what to recommend.

The audit to run this week

You can check your exposure without waiting for new Amazon reports. Pull your ASIN catalog and score each flagship product for agent-readiness. The fields that matter, per ASIN, are attribute completeness against the category's full attribute set, review recency and volume, price competitiveness, claim clarity in the title and bullets, and whether the detail page states use case and compatibility in plain language. Anything thin on attributes or stale on reviews is a product the assistant will skip.

We built a Claude skill that runs exactly this audit. Drop in your catalog export and it returns each ASIN's agent-readiness score, the missing attributes, the stale-review flags, and a prioritized fix list, sorted by the products carrying the most spend. Get the free skill.

The window is the point

Amazon is telling brands plainly that the keyword-and-detail-page funnel is being supplemented by an agent-led one, and that bidding, creative, and measurement will follow. Mid-size sellers have maybe two or three quarters before the largest brands absorb the agent shelf. The ones who instrument the shift and make their catalog legible now will be the products the assistant recommends. The ones who wait for a tidy click-replacement metric will get the answer when their rank delivers it, too late to act on.

Sources: Amazon Ads, what agentic shopping means for advertising (June 2026); Digiday, Amazon Alexa+ Agentic Ads coverage (June 18, 2026); Nova Data, Amazon Ads agentic shopping outlook (June 11, 2026); TechWyse, Amazon agentic shopping and Alexa advertising; Tinuiti 2026 AI Trends study.